Clean up Slack Bot formatting (#455)

backend/danswer/bots/slack/blocks.py

@@ -12,9 +12,7 @@ from danswer.bots.slack.utils import build_feedback_block_id
 from danswer.bots.slack.utils import translate_vespa_highlight_to_slack
 from danswer.configs.app_configs import DANSWER_BOT_NUM_DOCS_TO_DISPLAY
 from danswer.configs.app_configs import ENABLE_SLACK_DOC_FEEDBACK
-from danswer.configs.constants import DocumentSource
 from danswer.configs.constants import SearchFeedbackType
-from danswer.connectors.slack.utils import UserIdReplacer
 from danswer.direct_qa.interfaces import DanswerQuote
 from danswer.server.models import SearchDoc
 from danswer.utils.text_processing import replace_whitespaces_w_space
@@ -71,35 +69,6 @@ def build_doc_feedback_block(
     )
 
 
-def _build_custom_semantic_identifier(
-    semantic_identifier: str, match_str: str, source: str
-) -> str:
-    """
-    On slack, since we just show the semantic identifier rather than semantic + blurb, we need
-    to do some custom formatting to make sure the semantic identifier is unique and meaningful.
-    """
-    if source == DocumentSource.SLACK.value:
-        truncated_blurb = (
-            f"{match_str[:_MAX_BLURB_LEN]}..."
-            if len(match_str) > _MAX_BLURB_LEN
-            else match_str
-        )
-        # NOTE: removing tags so that we don't accidentally tag users in Slack +
-        # so that it can be used as part of a <link|text> link
-        truncated_blurb = UserIdReplacer.replace_tags_basic(truncated_blurb)
-        truncated_blurb = UserIdReplacer.replace_channels_basic(truncated_blurb)
-        truncated_blurb = UserIdReplacer.replace_special_mentions(truncated_blurb)
-        truncated_blurb = UserIdReplacer.replace_links(truncated_blurb)
-        # stop as soon as we see a newline, since these break the link
-        truncated_blurb = truncated_blurb.split("\n")[0]
-        if truncated_blurb:
-            return f"#{semantic_identifier}: {truncated_blurb}"
-        else:
-            return f"#{semantic_identifier}"
-
-    return semantic_identifier
-
-
 def build_documents_blocks(
     documents: list[SearchDoc],
     query_event_id: int,
@@ -144,40 +113,6 @@ def build_documents_blocks(
     return section_blocks
 
 
-def build_blurb_quotes_block(
-    quotes: list[DanswerQuote],
-) -> tuple[list[Block], list[str]]:
-    quote_lines: list[str] = []
-    doc_identifiers: list[str] = []
-    for quote in quotes:
-        doc_id = quote.document_id
-        doc_link = quote.link
-        doc_name = quote.semantic_identifier
-        if doc_link and doc_name and doc_id and doc_id not in doc_identifiers:
-            doc_identifiers.append(doc_id)
-            custom_semantic_identifier = _build_custom_semantic_identifier(
-                semantic_identifier=doc_name,
-                match_str=quote.blurb,
-                source=quote.source_type,
-            )
-            quote_lines.append(f"- <{doc_link}|{custom_semantic_identifier}>")
-
-    if not quote_lines:
-        return [], []
-
-    return (
-        [
-            SectionBlock(
-                fields=[
-                    "*Sources:*",
-                    *quote_lines,
-                ]
-            )
-        ],
-        doc_identifiers,
-    )
-
-
 def build_quotes_block(
     quotes: list[DanswerQuote],
 ) -> list[Block]:
@@ -206,19 +141,12 @@ def build_quotes_block(
         single_quote_str = "\n".join([f"```{q_str}```" for q_str in longest_quotes])
         link = doc_to_link[doc_id]
         sem_id = doc_to_sem_id[doc_id]
-        quote_lines.append(f"<{link}|{sem_id}>\n{single_quote_str}")
+        quote_lines.append(f"<{link}|{sem_id}>:\n{single_quote_str}")
 
     if not doc_to_quotes:
         return []
 
-    return [
-        SectionBlock(
-            fields=[
-                "*Relevant Snippets:*",
-                *quote_lines,
-            ]
-        )
-    ]
+    return [SectionBlock(text="*Relevant Snippets*\n" + "\n".join(quote_lines))]
 
 
 def build_qa_response_blocks(

backend/danswer/bots/slack/utils.py

@@ -12,6 +12,7 @@ from slack_sdk.models.metadata import Metadata
 from danswer.configs.app_configs import DANSWER_BOT_NUM_RETRIES
 from danswer.configs.constants import ID_SEPARATOR
 from danswer.connectors.slack.utils import make_slack_api_rate_limited
+from danswer.connectors.slack.utils import UserIdReplacer
 from danswer.utils.logger import setup_logger
 from danswer.utils.text_processing import replace_whitespaces_w_space
 
@@ -91,8 +92,7 @@ def decompose_block_id(block_id: str) -> tuple[int, str | None, int | None]:
 
 def translate_vespa_highlight_to_slack(match_strs: list[str], used_chars: int) -> str:
     def _replace_highlight(s: str) -> str:
-        s = re.sub(r"</hi>(?=\S)", "", s)
-        s = re.sub(r"(?<=\S)<hi>", "", s)
+        s = re.sub(r"(?<=[^\s])<hi>(.*?)</hi>", r"\1", s)
         s = s.replace("</hi>", "*").replace("<hi>", "*")
         return s
 
@@ -110,3 +110,11 @@ def translate_vespa_highlight_to_slack(match_strs: list[str], used_chars: int) -
         combined = combined[: remaining - 3] + "..."
 
     return combined
+
+
+def remove_slack_text_interactions(slack_str: str) -> str:
+    slack_str = UserIdReplacer.replace_tags_basic(slack_str)
+    slack_str = UserIdReplacer.replace_channels_basic(slack_str)
+    slack_str = UserIdReplacer.replace_special_mentions(slack_str)
+    slack_str = UserIdReplacer.replace_links(slack_str)
+    return slack_str

backend/danswer/datastores/vespa/store.py

@@ -196,9 +196,7 @@ def _index_vespa_chunks(
             headers: dict[str, str],
             fields: dict[str, Any],
         ) -> Response:
-            logger.debug(
-                f"Hitting URL '{url}', with headers '{headers}', with fields '{fields}'"
-            )
+            logger.debug(f'Indexing to URL "{url}"')
             res = requests.post(url, headers=headers, json={"fields": fields})
             try:
                 res.raise_for_status()